Meta-analysis of ordinal outcomes using individual patient data

Meta-analyses are being undertaken in an increasing diversity of diseases a nd conditions, some of which involve outcomes measured on an ordered catego rical scale. We consider methodology for undertaking a meta-analysis on ind ividual patient data for an ordinal response. The approach is based on the proportional odds model, in which the treatment effect is represented by th e log-odds ratio. A general framework is proposed for fixed and random effe ct models. Tests of the validity of the various assumptions made in the met a-analysis models, such as a global test of the assumption of proportional odds between treatments, are presented. The combination of studies with dif ferent definitions or numbers of response categories is discussed. The meth ods are illustrated on two data sets, in a classical framework using SAS an d MLn and in a Bayesian framework using BUGS. The relative merits of the th ree software packages for such meta-analyses are discussed. Copyright (C) 2 001 John Wiley & Sons, Ltd.